We included patients undergoing elective lung resection procedures using conventional and video-assisted thoracoscopic techniques in the Hospital de Messejana Dr. Carlos Alberto Studart Gomes in Fortaleza, Ceará State, Brazil, from July 2012 to November 2015. The Comitê de Ética em Pesquisa do hospital de Messejana (local Research Ethics Committee) approved the research protocol on May 16, 2011, under the number CEP 828/11.
Study population and inclusion criteria
The study population consisted of patients of both genders and of any race who were at least 18 years old. We excluded patients with at least one of the following characteristics: the impossibility of elective surgery, patient refusal, unlikelihood of 1-year follow-up after the surgical procedure, patients without troponin I measurements at the three-time predetermined points and patients with unstable coronary disease.
Study design
The present study is a prospective cohort study with planned endpoints and analysis. In the preoperative period, we obtained clinical data and surgical risk classification by the Multicenter study of perioperative evaluation for noncardiac surgeries in Brazil (EMAPO) and by the Detsky index of the ACP (American College of Physicians). The scores used in this study are described in detail elsewhere. 20-21
During the intraoperative period, we monitored patients for complications. A decrease in systolic blood pressure below 90 mmHg, a heart rate lower than 60 beats per minute, or the use of vasopressors or inotropes defined hemodynamic instability for this study. Intraoperatively use of bolus was not considered as haemodynamic instability criterion. Intraoperative arrhythmias were ventricular or supraventricular changes that occurred with hemodynamic instability, and that required intervention.
The postoperative management and discharge criteria were those defined in the routine guidelines of the hospital. To measure troponin I, blood samples were collected from all patients during the immediate postoperative period (POi) and the first and second PO. The analyses were performed using the Elecsys 2010 system from Roche®, 99th percentile of 0.16 ng/ml, and coefficient of variation <10% for values of 0.30 ng/ml. Considering the high specificity of troponin I, we choose an increase in TnI ≥0.16 ng/ml as the onset of myocardial injury. Patients were evaluated during the period of hospitalization and for 30 days after surgery for the presence of the following complications: acute pulmonary edema, stroke, acute myocardial infarction, cardiac arrest due to ventricular fibrillation, atrial fibrillation with hemodynamic instability, bleeding, pulmonary thromboembolism, respiratory infection, hypotension and death. Mortality within one year was assessed at outpatient visits or by phone call by an investigator blinded to the troponin levels.
Statistical analysis
For sample size calculation, we considered power of 80%, alpha 0.05, and estimated mortality rate of 15% and 2.5% in patients with and without elevated cardiac troponin I. The resulted sample size of our study was 158 patients.
We described the distribution of continuous variables as the mean and standard deviation, and the categorical variables as the relative frequency of the categories. The dependent variable of this study was survival for over one year. The independent variables were origin, gender, aetiology, age, weight, hypertension, coronary disease, diabetes mellitus, smoking habit, smoking load, previous radiotherapy and chemotherapy, functional capacity, surgical risk classification scores, type of surgery, use of video-assisted thoracoscopy, arrhythmia and intraoperative haemodynamic instability, blood transfusion, complications at 30 days and elevation of postoperative troponin I.
The normality of all numerical variables was tested using the Kolmogorov-Smirnov test. Parametric tests and regressive models checked the initial univariate analysis. We constructed Cox univariate regression models for each independent variable, and the outcome was death within one year. The variables that were significantly associated (p<0.05) with the outcome in the univariate analysis were input in a multivariate Cox regression model. The statistical program SPSS Inc., version 17.0, was used to perform the calculations.